Update README.md

README.md

@@ -84,8 +84,12 @@ This dataset only includes data.gouv.fr datasets that have at least a 1000 carac
 
 ### ✂️ 2. Text Chunking
 
-The `chunk_text` field was created by combining the `title`, `organization` name, `description`. The `description` was croped to a maximum length of +- 1000 caracters.
-The Langchain's `RecursiveCharacterTextSplitter` function was used to make these chunks. The parameters used are : 
+The `chunk_text` field was created by combining the `title`, `organization` name, `description`. 
+
+The `description` was croped to a maximum length of +- 1000 caracters.
+The Langchain's `RecursiveCharacterTextSplitter` function was used to crop the description. 
+
+The parameters used are : 
 - `chunk_size` = 1000 (in order to limit as much noise as possible)
 - `chunk_overlap` = 20
 - `length_function` = len